The Structure of Social Cognition: Precursors, Mechanisms, and Competencies

This is the authors' abstract. We don't add key points for this paper.

Social competence is an architecture rather than a unitary capacity, and the cognitive science of social behaviour remains fragmentedacross disciplines that do not always share a vocabulary. We propose a three-level decomposition, drawing on cognitive science,developmental psychology, philosophy of mind, psycholinguistics, social ethology and AI research. Seven candidate socio-cognitiveprecursors (embodied, pre-representational dispositions such as animacy detection, gaze following and emotional contagion) sitbelow seven socio-cognitive mechanisms (goal inference, mentalizing, mutual modelling, shared intentionality, meta-cognitive socialmodelling, norm representation, reciprocity tracking), which compose into thirty-eight domain competencies. The mechanisms areindividuated by dissociability and ordered by their dependencies, with the evidence stated for each boundary. That ordering carriesour main structural claim: goal inference, not mentalizing, is the foundational mechanism, and coordination, joint commitment andcommon ground (short of its epistemic form) do not require belief attribution. This departs from the theory-of-mind-centred accountsthe field has organised itself around. We also discuss two factors that are not specific to social cognition, yet play a fundamental role init: situatedness, and central coherence, which we argue is a parameter on belief inference rather than a socio-cognitive mechanismper-se. Finally, we compile a catalogue of experimental paradigms, including matched “fine-cuts” task pairs that separate surface socialperformance from socio-cognitive competence, and set out how the framework could be tested by socio-cognitive ablation.